PolitiCause: An Annotation Scheme and Corpus for Causality in Political Texts

Abstract

An Annotation Scheme and Corpus for Causality in Political Text. PolitiCAUSE is a corpus of political texts annotated for causality. We provide two types of information: (1) whether a sentence contains a causal relation or not (2) the spans of text that correspond to the cause and effect components of the causal relation. The dataset is available in two ways: (1) As a full dataset containing all annotations and statistics for 55,754 annotation instances. (2) As a train, validation and test splits containing the text and the label of 17,780 unique sentences. We benchmarked the dataset using three transformer-based classification models, the models achieve a moderate performance on the dataset, with a MCC score of 0.62. PolitiCAUSE is a valuable resource for studying causality in texts, especially in the domain of political discourse.